USF Libraries
USF Digital Collections

Design and testing of a motion controlled gait enhancing mobile shoe (gems) for rehabilitation

MISSING IMAGE

Material Information

Title:
Design and testing of a motion controlled gait enhancing mobile shoe (gems) for rehabilitation
Physical Description:
Book
Language:
English
Creator:
Handzic, Ismet
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla
Publication Date:

Subjects

Subjects / Keywords:
Archimedean Spiral
Foot Haptics
Gait Adaptation
Gear Mechanics
Human Gait Rehabilitation
Dissertations, Academic -- Mechanical Engineering -- Masters -- USF   ( lcsh )
Genre:
bibliography   ( marcgt )
non-fiction   ( marcgt )

Notes

Summary:
ABSTRACT: Persons suffering central nervous system damage, such as a stroke, coma patients, or individuals that have suffered damage to the spinal cord, brainstem, cerebellum, and motor cortex, sometimes develop an asymmetric walking pattern where one leg does not fully swing backward. This uneven gait hinders these individuals in properly and efficiently moving through everyday life. Previous research in humans and various animals has introduced a split belt treadmill to analyze possible rehabilitation, which can recreate a correct gait pattern by altering the speed of each track. Gait adaptation was achieved by having the split belt treadmill move each leg at a different velocity relative to the ground and thus forcing a symmetric gait. Test subjects‟ gait would adapt to the speeds and a normal gait pattern could be conditioned while on the split belt treadmill. However, after short trials, individuals were unable to neurologically store these feed-forward walking patterns once walking over ground. Also, test subjects would have difficulty adapting their learned walking gait over different walking environments. The gait enhancing mobile shoe (GEMS) makes it possible to adjust an asymmetric walking gait so that both legs move at a relatively symmetric speed over ground. It alters the wearers walking gait by forcing each foot backwards during the stance phase, operating solely by mechanical motion, transferring the wearer‟s downward force into a horizontal backwards motion. Recreating the split belt treadmill effect over ground by using the GEMS will potentially enable me to test the long term effects of a corrected gait, which is impossible using a split belt treadmill. A previous prototype of the GEMS 1 successfully generated a split belt treadmill walking pattern, but had various drawbacks, such as variable motion from step to step. My new design of this rehabilitation shoe promises to alter the user‟s gait as a split belt treadmill does, and to be mechanically stable operating without any external power sources. I designed and constructed a new motion controlled gait enhancing mobile shoe that improves the previous version‟s drawbacks. While mimicking the asymmetric gait motion experienced on a split-belt treadmill, this version of the GEMS has motion that is continuous, smooth, and regulated with on-board electronics. An interesting aspect of this new design is the Archimedean spiral wheel shape that redirects the wearer‟s downward force into a horizontal backward motion. The design is passive and does not utilize any motors and actuators. Its motion is only regulated by a small magnetic particle brake. Initial tests show the shoe operates as desired, but further experimentation is needed to evaluate the long-term after-effects.
Thesis:
Thesis (M.S.M.E.)--University of South Florida, 2011.
Bibliography:
Includes bibliographical references.
System Details:
Mode of access: World Wide Web.
System Details:
System requirements: World Wide Web browser and PDF reader.
Statement of Responsibility:
by Ismet Handzic.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains 110 pages.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
usfldc doi - E14-SFE0004862
usfldc handle - e14.4862
System ID:
SFS0028126:00001


This item is only available as the following downloads:


Full Text
xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam 22 Ka 4500
controlfield tag 007 cr-bnu---uuuuu
008 s2011 flu ob 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0004862
035
(OCoLC)
040
FHM
c FHM
049
FHMM
090
XX9999 (Online)
1 100
Handzic, Ismet
0 245
Design and testing of a motion controlled gait enhancing mobile shoe (gems) for rehabilitation
h [electronic resource] /
by Ismet Handzic.
260
[Tampa, Fla] :
b University of South Florida,
2011.
500
Title from PDF of title page.
Document formatted into pages; contains 110 pages.
502
Thesis
(M.S.M.E.)--University of South Florida, 2011.
504
Includes bibliographical references.
516
Text (Electronic thesis) in PDF format.
520
ABSTRACT: Persons suffering central nervous system damage, such as a stroke, coma patients, or individuals that have suffered damage to the spinal cord, brainstem, cerebellum, and motor cortex, sometimes develop an asymmetric walking pattern where one leg does not fully swing backward. This uneven gait hinders these individuals in properly and efficiently moving through everyday life. Previous research in humans and various animals has introduced a split belt treadmill to analyze possible rehabilitation, which can recreate a correct gait pattern by altering the speed of each track. Gait adaptation was achieved by having the split belt treadmill move each leg at a different velocity relative to the ground and thus forcing a symmetric gait. Test subjects‟ gait would adapt to the speeds and a normal gait pattern could be conditioned while on the split belt treadmill. However, after short trials, individuals were unable to neurologically store these feed-forward walking patterns once walking over ground. Also, test subjects would have difficulty adapting their learned walking gait over different walking environments. The gait enhancing mobile shoe (GEMS) makes it possible to adjust an asymmetric walking gait so that both legs move at a relatively symmetric speed over ground. It alters the wearers walking gait by forcing each foot backwards during the stance phase, operating solely by mechanical motion, transferring the wearer‟s downward force into a horizontal backwards motion. Recreating the split belt treadmill effect over ground by using the GEMS will potentially enable me to test the long term effects of a corrected gait, which is impossible using a split belt treadmill. A previous prototype of the GEMS [1] successfully generated a split belt treadmill walking pattern, but had various drawbacks, such as variable motion from step to step. My new design of this rehabilitation shoe promises to alter the user‟s gait as a split belt treadmill does, and to be mechanically stable operating without any external power sources. I designed and constructed a new motion controlled gait enhancing mobile shoe that improves the previous version‟s drawbacks. While mimicking the asymmetric gait motion experienced on a split-belt treadmill, this version of the GEMS has motion that is continuous, smooth, and regulated with on-board electronics. An interesting aspect of this new design is the Archimedean spiral wheel shape that redirects the wearer‟s downward force into a horizontal backward motion. The design is passive and does not utilize any motors and actuators. Its motion is only regulated by a small magnetic particle brake. Initial tests show the shoe operates as desired, but further experimentation is needed to evaluate the long-term after-effects.
538
Mode of access: World Wide Web.
System requirements: World Wide Web browser and PDF reader.
590
Advisor:
Reed, Kyle B.
653
Archimedean Spiral
Foot Haptics
Gait Adaptation
Gear Mechanics
Human Gait Rehabilitation
690
Dissertations, Academic
z USF
x Mechanical Engineering
Masters.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.4862



PAGE 1

Design and Testing of a Motion Controlled Gait Enhancing Mobile Shoe (GEMS) for Rehabilitation by Ismet Handzic A thesis submitted in partial fulfillment o f the requirements for the degree of Master of Science in Mechanical Engineering Dep artment of Mechanical Engineering College of Engineering University of South Florida Major Professor: Kyle B. Reed Ph.D Rajiv Dubey, Ph.D Craig Lusk, Ph.D Date of Approval : March 22 2011 Keywords: Human Gait Rehabilitation, Gait Adaptation, Foot Haptics, Gear Mechanics, Archimedean Spiral Copyright 2011 Ismet Handzic

PAGE 2

Acknowledgements To my advisor, Dr. Kyle B Reed, for his knowledge in the field of mechanical engineering, electrical circuits, and mechanical controls. To Ron Rizzo of Western Kentucky University for his knowledge in microcontrollers. To the machine shop for their excellent machining and assembly of various parts of this thesis project. This work is funded by a NIH grant (1R21HD066200 01).

PAGE 3

i Table of Contents List of Tables ................................ ................................ ................................ ................ iii List of Figures ................................ ................................ ................................ .............. iv Abstract ................................ ................................ ................................ ...................... v i i i Chapter 1: Introduction ................................ ................................ ................................ 1 1.1 Project Motivation ................................ ................................ .................. 1 1.2 Design Goals ................................ ................................ .......................... 3 1.3 Section Overview ................................ ................................ ................... 7 Chapter 2: Background ................................ ................................ ................................ 9 2.1 Human Walking Gait ................................ ................................ .............. 9 2.2 Gait Rehabilitation ................................ ................................ ............... 16 2.3 GEMS and Context Awareness ................................ ............................. 21 2.4 Previous GEMS M odel ................................ ................................ ......... 22 Chapter 3: Design and Assembly ................................ ................................ ............... 27 3 .1 Wheel Shape ................................ ................................ ........................ 28 3 .2 Gear Train and Magnetic Particle Brake ................................ ............... 34 3. 2 .1 Magnetic Particle Brake ................................ ...................... 3 6 3. 2 2 Gear Train ................................ ................................ ........... 3 8 3 .3 Wheel Re set Mechanism ................................ ................................ ...... 41 3 .4 Electronics ................................ ................................ ............................ 4 7 3.4.1 Microcontroller ................................ ................................ ..... 48 3. 4.2 Accelerometer ................................ ................................ ...... 50 3.4.3 Potentiometer ................................ ................................ ....... 52 3.4.4 Battery Pack ................................ ................................ ......... 54 3 .5 GEMS Frame ................................ ................................ ....................... 56 3 .6 Shoe Straps ................................ ................................ ........................... 57 3 .7 Opposite Leg Support Platform ................................ ............................ 58 3 .8 CAD Model Verification ................................ ................................ ...... 59 Chapter 4 : Results ................................ ................................ ................................ ...... 61 Chapter 5: Future Work ................................ ................................ ............................. 67

PAGE 4

ii Chapter 6 : Conclu sions ................................ ................................ .............................. 69 References ................................ ................................ ................................ ................... 71 Appendices ................................ ................................ ................................ .................. 75 Appendix A GEMS Electrical Diagram ................................ ......................... 76 Appendix B Pugh Analysis: GEMS Frame Material ................................ ..... 77 Appendix C Archimedean Spiral W heel Shape Selection Tool ...................... 78 Appendix D Specification Sheets ................................ ................................ .. 79 Appendix E Dimensions and Drawings ................................ ......................... 87 Appendix F Microcontroller Cod e ................................ ................................ 98

PAGE 5

iii List of Tables Table 1: Wheel shape parameters chosen for the new GEMS ................................ ... 3 1 Table 2: Extension s pring p roperties ................................ ................................ ........ 4 5 Table 3: Power requirement for GEMS components ................................ ................. 54 Table 4: New and old GEMS compared by criteria ................................ ................... 66 Table D1: BS2p24 Microcontroller specification ................................ ........................ 8 0

PAGE 6

iv List of Figures Figure 1: Adult h uman w alking g ait c ycle ................................ ................................ 9 Figure 2: Time d uration for each leg in gait cycle ................................ ................... 10 Figure 3: Vicon motion camera ................................ ................................ .............. 12 Figure 4: Conductive mat ................................ ................................ ....................... 12 Figure 5: The horizontal (Fx) and vertical (Fz) forces change throughout the gait cycle. ................................ ................................ ................................ 1 4 Figure 6: Conventional Escalator ................................ ................................ ............ 15 Figure 7: Con ventional Split belt Treadmill ................................ ........................... 18 Figure 8: Rehabilitation method for asymmetric gait based on exaggeration where the GEMS is worn on lagging leg ................................ .................. 20 Figure 9: Rehabilitation method for asymmetric gait based on compensation where the GEMS is worn on healthy l eg ................................ ................. 2 1 Figure 10: Initial existing GEMS prototype ................................ .............................. 2 3 Figure 11: Rear mechanisms that cause backward moti applied ................................ ................................ ................................ .... 2 4 Figure 12: GEMS motion throughout the stance phase ................................ ............. 2 5 Figure 13: The GEMS wheel shape is designed so that when normal vertical force is applied it behaves as in rolling down a slope ............................... 28 Figure 14: The GEMS wheel sha pe redirects vertical user weight force into a horizontal ground reaction applying a torque at the wheel axle creating forward progression ................................ ................................ ... 30 Figure 15: GEMS Wheels shape length, L, and radius, R, parameters ....................... 30 Figure 16: Side view of actual GEMS wheel mated to axle ................................ ....... 33

PAGE 7

v Figure 17: Side view of the gear train of the GEMS including magnetic particle brake ................................ ................................ ................................ ....... 35 Figure 18: 1lb in magnetic particle brake utilized in the GEMS design ..................... 36 Figu re 19: Non I nverting op amp circuit used to output 0 24 to the magnetic particle brake ................................ ................................ ........................... 37 Figure 20 : GEMS gear train ................................ ................................ ..................... 39 Figure 21 : Front m iter gear aluminum bracket ................................ .......................... 40 Figure 22 : Aluminum plate on top of the GEMS frame cover preventing unmashing of m iter gear set ................................ ................................ ..... 40 Figure 23: Top view of the GEMS showing the wheel reset mechanism ................. 41 Figure 24: To estimate the spring force need to reset the GEMS, a simple setup wa s used ................................ ................................ ................................ .. 43 Figure 25: The extension springs are pre tensioned by a nylon string attached to the frame ................................ ................................ ................................ 44 Figure 26: Aluminum reset pulley, redirect pulley, and doubled nylon string ............ 45 Figure 27: Spring to nylon string bracket (left) and spring to fra me bracket (right) ................................ ................................ ................................ ...... 46 Figure 28: Parallax BS2p24 microcontroller integrated into the GEMS .................... 48 Figure 29: Circuit board for GEMS ................................ ................................ .......... 50 Figure 30: Position of accelerometer inside the GEMS ................................ ............. 5 1 Figure 31: Circuit board for the accelerometer ................................ .......................... 5 2 Figure 32: Potentiometer attached to the gear train shaft ................................ ........... 5 3 Figure 33: GEMS battery pack ................................ ................................ ................. 55 Figure 34: GEMS Frame and bottom of frame Cover ................................ ............... 57 Figure 35: The GEMS straps are designed after traditional sandals ........................... 58 Figure 36: Platform used to compensa te for the height of the GEMS ........................ 59

PAGE 8

vi Figure 37: GEMS SolidWorks 3D CAD model ................................ ........................ 60 Figure 38: ................................ .................. 61 Figure 39: Interior of complete GEMS ................................ ................................ ..... 62 Figure 40: GEMS stepping motion ................................ ................................ ........... 6 3 Figure A1 : GEMS Electrical Diagram ................................ ................................ ...... 76 Figure B1: Pugh Analysis: GEMS frame material ................................ ..................... 77 Figure C1: Archimedean Spi ral wheel shape selection tool ................................ ...... 78 Figure D1: BS2p24 module schematic ................................ ................................ ..... 79 Figure D2: Accelerometer specifications ................................ ................................ ... 81 Figure D3: Magnetic particle brake specifications ................................ ..................... 82 Figure D4: Op Amp specifications ................................ ................................ ............ 83 Figure D5: Sprocket conne cting front and back GEMS axle ................................ ...... 84 Figure D6: Chain connecting front and back GEMS axl e ................................ .......... 85 Figure D7: Miter gear in GEMS gear train ................................ ................................ 86 Figure E1: GEMS assembly ................................ ................................ ...................... 87 Figure E 2: Open GEMS assembly top view ................................ .............................. 88 Fi gure E3: GEMS wheel ................................ ................................ ........................... 89 Figure E4: GEMS frame bottom cover ................................ ................................ ...... 90 Figure E5: GEMS frame side ................................ ................................ .................... 91 Figure E6: GEMS frame top cove r ................................ ................................ ............ 92 Figure E7: Sixty tooth gear in gear train ................................ ................................ .... 93 Figure E8: Miter gear bracke t ................................ ................................ ................... 94 Figure E9 : Potentiometer bracket ................................ ................................ .............. 95

PAGE 9

vii Figure E10: Reset mechanism redirect pulley bracket ................................ ................. 96 Figure E11: Small reset mechanism pulley ................................ ................................ 97

PAGE 10

viii Abstract Persons suffering central nervous system damage, such as a stroke, coma patients, or individua ls that have suffered damage to the spinal cord, brainstem, cerebellum, and motor cortex, sometimes develop an asymmetric walking pattern where one leg does not fully swing backward. This uneven gait hinders these individuals in properly and efficiently mov ing through everyday life. Previous research in humans and various animals has introduced a split belt treadmill to analyze possible rehabilitation, which can recreate a correct gait pattern by altering the speed of each track. Gait adaptation was a chieved by having the split belt treadmill move each leg at a different velocity relative to the ground and thus forcing a symmetric could be condition ed while on the split b elt treadmill. However, after short trials, individuals were unable to neurologically store these feed forward walking patter n s once walking over ground Also, test subjects would have difficulty adapting their learned walking gait over different walking environments. The gait enhancing mobile shoe (GEMS) makes it possible to adjust an asymmetric walking gait so that both legs move at a relatively symmetric speed over ground. It alters the wearers walking gait by forcing each foot backwards during the s force into a horizontal backwards motion. Recreating the split belt treadmill effect over

PAGE 11

ix ground by using the GEMS will potentially enable me to test the long term effec ts of a corrected gait, which is impossible using a split belt treadmill. A previous prototype of the GEMS [1] successfully generated a split belt treadmill walking pattern, but had various drawbacks, such as variable motion from step to step. My new desi split belt treadmill does, and to be mechanically stable operating wit hout any external power sources. I designed and constructed a new motion controlled gait enhancing mobile shoe that improves motion experienced on a split belt treadmill, this version of the GEMS has motion that is continuous, smooth, and regulated with on board electronics. An interesting aspe ct of downward force into a horizontal backward motion. The design is passive and does not utilize any motors and actuators. Its motion is only regulated by a small magneti c particle brake. Initial tests show the shoe operates as desired, but f urther experimentation is needed to evaluate the long term after effects.

PAGE 12

1 Chapter 1 : Introduction 1. 1 : P r o j e c t M o t i v a t i o n While walking all healthy humans have a relatively similar walking pattern (gait). Each leg performs the same movements in a cycle 180 degrees out of sync from each other; this repetition of leg and foot movements is called the gait cycle. It involves two parts, or one stride, and it is divided into the stance phase and the swing phase. The stance phase is the period during which the foot is in contact with the ground and the swing phase is the period of the gait cycle when the foot is swinging forward. Stroke patients [2] coma patient s, and/or people with cen tral nervous system damage [3, 2, 4] sometimes develop an asymmetric walking gait preventing them from move ing around normally in everyday life. Such individuals are unable to continuously perform a correct symmetric gait cycle. In these hemiplegic patients, one leg lags the other, not traveling far enough backwards to effectively push the individual forward during walking. This handicap creates an asymmetry that can result in a worsening asymmetry [ 22] and so strain Studies, where the strategies used on hemiplegic subjects to adapt their walking pattern to a velocity dependent resistance applied against hip and knee movements, showed there are two basic ways the human body alters its n ormal walking gait [3]. One is by a feedbac k driven, or reactive adaptation that occurs in response to a sudden external perturbation. This type of gait adaptation does not require any previous training. The second is a feed forward driven adaptation [3 ], where training is required to

PAGE 13

2 neurologica lly store certain movement patterns into muscle memory. This type of gait conditioning produces longer aftereffects. It was shown that using a split belt treadmill with asymmetric belt speed velocity ratios [4, 2, 5] allowed individuals to adap t their walking gait to the asymmetric belt speeds showing trained aftereffects. When the split belt treadmill is returned back to a 1:1 ratio, the individual walked with a trained and altered gait. This altere d walking pattern, however, vanishes after a short period of time walking on a 1:1 ratio split belt This type of gait alteration is a feed forward gait adaptation. Riesman et al. [2] suggest that long term ef fects need to be studied in real world situations. These long term effects of correcting an asymmetric gait can better be achieved with a mobile shoe, which a test subject would wear for an extensive period of ti me in multiple environments. This concept has evolved into the Gait Enhancing Mobile Shoe (GEMS). The portable GEMS imitates the same relative foot motion experienced in previous split belt treadmill gait rehabilitation methods, but while walking over gr ound. In other words, as one leg travels one complete step, the other leg only covers a fraction of the distance covered by the first leg, hence, mimicking an asymmetric split belt treadmill. Other advantages of such a portable rehabilitation device are t hat it can be worn in extended period of time, thus the corrected gait is predicted to persist longer than the gait correction from a split belt treadmill. Moreo ver, the ability to wear the GEMS for longer periods of time increases the probability of producing positive gait rehabilitation effects.

PAGE 14

3 There are two basic ways the GEMS rehabilitates asymmetric walking patterns. The first method utilized by the GEMS co rrects an asymmetric gait by letting the test subject wear the shoe on the strong leg making it swing back past the stance phase and letting the individual toe off (push off) with that leg. In a way it hinders full forward progression of the healthy leg a s a slippery or icy surface would do. As a result both legs will propagate the individual forward by similar distances. This is compensation, but no permanent rehabilitation will come from this method. However, this compensation method will assist individ uals to walk with a symmetric gait. The second method is letting individual wear the shoe on the weak leg and so limiting the forward motion of the weak leg. This motivates the individual to lengthen the forward distance to the point where the initial hee l contact position would be. The GEMS also pushes the leg backwards forcing the weak leg to toe off properly and so recreating a normal gait. Considering past hemiplegic research, it is hypothesized that prolonged wearing of the GEMS will have positive a fter effects helping individuals with asymmetric walking patterns adapt a more normal walking gait when wearing the GEMS over a longer period. 1.2 : Design Goals It is my intention to create a device that adds to t he rehabilitation concept previously outlined in section 1.1. In general terms, it is my intention to construct a mobile shoe that mimics the kinematics of past split belt research while eliminating the context awareness problem d uring gait adaptation.

PAGE 15

4 Unlike the previous GEMS, this model is to be fully motion controlled in order to make it more versatile to various subjects and walking conditions. While the previous model was solely based on a mechanical design, the new design is going to utilize electrical components to control and regulate the motion of the shoe. The new version of the GEMS is controlled by an on board microprocessor. This microprocessor takes inputs from an accelerometer and potentiometer to recognize where in the gait cycle the GEMS is located. This microprocessor also will regulate the resistance of the shoe as it pushes rotation and a gear train that reduces the input torq ue from the wheels to the brake. All the electrical components are to be powered by a battery pack worn around the waist. Below are the ultimate design requirements for the GEMS. 1. Total weight : The total weight of the shoe should be no more than 2.2lb (1 k g), which is only slightly more than a typical tennis shoe, which weighs roughly1.65lb (0.75 kg). The first prototype weighs 2.43lb (1.1 kg) and could easily have some of the weight in the frame reduced. The second prototype weighs 4.4lb (2 kg), which will need significant optimization in order to reduce the weight to the desired amount. 2. Strength : The shoe should be capable of fully supporting the dynamic forces from a 115 kg person. This means a 115 kg wearer can stand or walk on it without failure. This d oes not mean that the shoe can stop the horizontal motion at any point prior to the toe off stage. The first prototype meets this requirement, but the second one becomes unreliable when a person weighing more than 90 kg uses the shoe.

PAGE 16

5 3. Generated motion: The total generated horizontal difference between the motions of the two feet during stance should be at least 30 cm. This means one foot can go forward 15 cm and one foot can go backward 15 cm, which would enable a total difference of 30 cm. Both prototypes meet this requirement. The first prototype can generate a backward motion of 25 35 cm, depending on the step. The second prototype can generate a 15.2 cm motion, so with a similar shoe on the other foot it would meet this requirement. 4. Consist ent motion : The horizontal motion generated by the shoe should be consistent on every step, at most the variability can be 5%. It is currently unclear what amount of variability would still permit an adaptation. However, it is clear that less is better. T he first prototype does not meet this requirement. In initial testing, it appears that the second prototype meets this requirement, but further testing is needed. 5. Portability : The shoe should be completely portable, with no external cables. If necessary, a small battery pack attached to the leg or hip would be acceptable. Both prototypes meet this requirement. 6. Time to recharge : The wearer should be able to walk continuously on the shoe for at least 1.5 hour. The first prototype meets this requirement. The s econd prototype has not been tested to this extent. 7. Size (height) : The height of the shoe (i.e., from the ground to the bottom of the

PAGE 17

6 Some height is necessary for the actual shoe components and also for redirection of forces based on a small downward motion being transferred to the required backward or forward motion. Neither prototype meets this requirement at the current time. Based on the size of the current shoes, optimizi ng the forces on the wheels and springs as well as determining the tolerances for the internal parts will allow this requirement to be met on future versions. 8. Size (width) : The width of the shoe should be similar to that of a typical tennis shoe. In no cas e should the shoe protrude 1 inch more to the inside of the leg than a typical tennis shoe does. A protrusion of 2 inches or less to the outside of the foot is acceptable since this will not interfere with walking. Both prototypes meet this requirement. 9. Si ze (length) : The length of the shoe should be similar to that of a typical tennis shoe. A longer shoe is acceptable as long as there is no interference with either the typical toe off nor heel contact events. Both prototypes meet this requirement. 10. Cost eff ectiveness : The shoe should be cost effective to manufacture. Assuming reasonable economies of scale, the shoe should be able to be produced for less than $500. Neither of the first two prototypes meet this goal as they were both custom made and have not b een optimized for manufacturing.

PAGE 18

7 11. Shoe Progression: The GEMS should utilize a special shaped wheel. This wheel has a form of constant radius change (Archimedean Spiral). It transfers the users vertical weight into a horizontal force. This wheel shape s hould allow a total push distance of the shoe of at least 6.5in (15cm). 12. Straps: The GEMS design should have straps so a user can fasten their everyday shoe on top of it. The straps should be designed in a way to where there is minimal movement between t 13. Opposite Foot Support Platform: A platform for the opposite leg should be made to compensate to any height difference created by the GEMS between both legs. This second support platform has identical variat ion in dimension and weight to reduce unnecessary inconsistencies during operation of the GEMS. 1.3 : Section Overview In the foregoing sections I will be describ e a clear and complete background to the work presented in this manuscript. The backg ro und covers the kinematics and kinetics of the normal and abnormal human gait and a hemiplegic rehabilitation method used to improve gait asymmetry The background chapter also covers a summery of the proposed gait enhancing mobile shoe concept and as it r elates to context awareness. To complete the background to this project, the previous GEMS i abilities, strengths and weaknesses.

PAGE 19

8 A detailed description of 3 of this manu script. This design and assembly Chapter explains the individual components of the GEMS, their relevance in a global view of the project and their detailed setup and layout. In Chapter 4 the results of the project is presented in comparison to the init ial design goals presented in Section 1.3 Each criteria is weighted against the resulting valued obtained by the new GEMS design. While this project shows promising results, room for future improvement and optimization is plentiful. The future work c hapter discretely describes each area of improvement and elaborates on future possibilities.

PAGE 20

9 Chapter 2 : Background 2.1 : Human Walking Gait It is important to describe the kinematics and other details of normal human gait before di scussing the applications of the GEMS and its rehabilitation methods. Walking walking pattern movement is rhythmic and symmetric in nature with a constant velocity forward pr ogression. The forward and backward walking motions are exactly similar in kinematics but differ in EMG activity in various leg and foot muscles [6]. For modern life mobility through the process of walking is a crucial activity of dai ly life A major abnormality such as a hemiplegic gait can greatly hinder an individual to eff iciently go through the process of everyday activities. A scientific poll has shown walking to be the most significant ability during recover y in stroke patients [ 23] The kinematics of the human gait cycle can be divided into two distinct phases: The stance phase and the swing phase [7]. Furthermore, these two gait cycle phases can be broken down into seven gait cycle instances shown in Figu re 1. When walking these gait cycle sub phases are exactly mirrored 180 degrees out of phase for each leg. Figure 1 : Adult h uman w alking g ait c ycle

PAGE 21

10 Although the human gait has the same characteristics in all healthy individuals, there are slight variations in this gait in every individual. Variations in healthy human gait stems from the fact that each individual learns this pattern early in life [ 24]. It is also universally known that many different variables affect the walking pattern of an individual; such variables includ e but are not limited to shoe cushioning, shoe height, y listed, though there is an average human gait to describe the details about the pattern, everyone has their own variation in walking. Figure 2 : Time duration for each leg in gait cycle Figure 2 shows the time spent by each foot in stance and swing phase. Notice that there is a time where both feet are contacting the ground at the same time; this is referred r pace the double support time

PAGE 22

11 decreases and finally disappears as the individual starts running. The step length is the distance where the left heel strikes the ground to the point where the right heel strikes the ground. Granat et al. [ 25] use a special ly designed shoe insole with switches to divide a hemiplegic gait into subcategories. This study adds an interesting aspect of a hemiplegic gait cycle called scuffing. Scuffing is described as any foot contact during the swing phase. In other words, as the foot toes off the ground it initiates into the swing phase W hile the foot swings from this toe off to ward heel contact it makes contact with the ground. Different gait analysis systems have been developed to further analyze human gait, these analy sis tools range from motion cameras for detailed kinetic and kinematic gait analysis to custom or specialized devices for foot pressure and temporal analysis [ 25] Although there are some non technological ways such as using foot prints and stop watches, gait analysis systems which can track distance and time of the gait cycle (temporal and special gait measurements) and are categ orized as follows: Motion Cameras Motion cameras come in a variety and range from more th a n one camera capturing the same mo tion from different angles to form one motion picture to infrared cameras that utilize reflective nodes placed on a moving object. 26]

PAGE 23

12 Figure 3 : Vicon m otion c amera [27 ] Conductive mat A simple device that it self changes electrical resistivity in order to map static forces/pressure applied to its surface. These static force or pressure is then turned into electrical signals which in turn become two dimensional pictures available for analysis Figure 4 : Condu ctive mat [28 ]

PAGE 24

13 Resistive walkway This type of system utilizes the change in resistivity of a resistance grid laid across the floor. As a subject walks across the floor, the system measures the change in resistance and so extract s time and distance values 29, 30] Switch Walkway This category can be subdivided into two systems: Little switches imbedded inside the soles of a mobile shoe and switches that are distributed across a walkway. During walking as the individ ual activates a switch, the system can determine the timing and pressure distribution of each step. This is a relatively low cost method of determining time and distance measurements of the gait cycle. Considering the kinetics of the gait cycle, Figure 5 shows the horizontal and vertical forces during the stance phase in an adult walking gait starting at heel contact. These reaction forces are exactly identical in both legs for a healthy human gait.

PAGE 25

14 Figure 5 : The horizontal (Fx) and vertical (Fz) for ces change throughout the gait cycle. The GEMS uses these changing forces to alter gait patterns for rehabilitation Notice that during the stance phase, a slightly fluctuating 800N vertical force is applied, while horizontal reaction forces of up to 100 N are experienced. Also notice that at the 30% gait cycle mark, the horizontal reaction force switches from pushing the leg backward to pushing the leg forward. This shows that the leg is actually slowing the person down during the first 30% and is pushi ng the person forward during the last 70% of the gait cycle The peak of the horizontal forward push occurs right before heel rise and toe off after which the swing phase is initiated. While the kinetics and kinematics of the human gait are ea sily measureable, it also has a psychological side to it. Walking involves context awareness [9], or location awareness. Context awar

PAGE 26

15 perturbations to the physical body while preparing and adjusting fo r such disturbances. The human body is a learning machine, which adapts to external perturbations with conscious and unconscious reaction and balancing forces. A good example of this gait context awareness is when someone is about to step on to a non moving escalator. Their body automatically adjusts muscle tension and aligns its center of gravity anticipating escalator movement. Similarly, this can be observed by analyzing the preplanned trajectories when reaching for objects a known distance away [10, 11]. During walking if a repeated permutation is anticipated, an individual adapts their walking pattern in an unconscious and automa tic manner to expend the least amount of energy expenditure Figure 6 : Conventional Escalator [41 ] Abnormalities in the human gait are numerous and are a whole topic on their own. For the relevance to this manuscript I focus on the anatomy of the asymmetric walking pattern adapted by persons who have experienced central nervous system damage, suc h as stroke [ 31, 32], Parkinson disease [33 ] coma patients, or individuals that have suffered damage to the spinal cord, brainstem, cerebellum [ 31] and motor cortex. This division in

PAGE 27

16 abnormal gait I will refer to as asymmetric gait or hemiplegic gait. Li ttle is known about the direct causes of gait asymmetries, however some studies in various animals [ 34, 35] suggest that abnormalities in rhythmic patterns such as walking stems in the auto activity of localized networks of neurons or central pattern gener system. Also, studies in infants and adults [ 36, 37] lead to the conclusion that there are central pattern generators responsible for left right leg pattern movement. While the essential neurological causes of hemiplegic gait are still uncertain, the physical anatomy of a hemiplegic gait is clear and measurable. Using a wearable planer footprint analysis system, Gavira et al. [ 38] showed that hemiplegic patients spend less time and rely less on their lagging foot for supp ort. It was also shown that the stride length was much shorter, walking velocity much lower, step duration longer, and reactive force values and temporal correspondences significantly different in the midfoot and forefoot. 2.2 : Gait Rehabilitation A normal gait is very adaptive in nature and can be altered over a conscious short term feedback reaction type movement when sudden perturbations are introduced, such as an individual slightly tripping over a folded rug and recovering. Normal gait also has been shown to be altered in a more unconscious fashion with longer feed forward learned type movements when continuous external physical stimuli is applied onto each limb while walking [3] or when walking with an uneven belt speed ratio on a split belt tr eadmill [2, 21]. The repetitive asymmetric stimulus on a symmetric gait over a longer period of time has been shown to have altering aftereffects.

PAGE 28

17 Not only has it been shown that the adult human gait can be altered to automatically perform a learned mov ement, but studies in animals and infants [5] have shown similar results in ferrets [12] spinalized cats [13], rats [14], and frogs [15]. These studies also show that damage to the cerebella in animals hinders the ability for a feed forward learned type o f adaption in gait but do not affect the feedback reaction type alteration in walking gait [17] suggesting that the feed forward gait ad aptation is controlled by the cerebella. Reisman et al. [2] showed this principle by setting up a split b elt treadmill as shown in Figure 7 where the track velocity ratio was forced to 2:1. Hemiplegic patients were conditioned with this asymmetric track velocity ratio for fifteen minutes during which the subjects gait adapted to a more symmetric gait. As t he split belt treadmill was switched back to a 1:1 track velocity ratio a symmetric gait after effect was observed. Although there was a positive rehabilitation effect using this split belt treadmill gait conditioning method, the aftereffect was short liv ed lasting only for a couple of seconds.

PAGE 29

18 Figure 7 : Conventional Split belt Treadmill [ 39] It is suggested that the long term aspects of this effect should be studied. These long term split belt treadmill effects are of course impossible to study wit h a split belt treadmill. A test subject can only be held on a split belt treadmill for a short period of time. Hence, a mobile device such as the gait enhancing mobile shoe (GEMS) can be utilized to observe such long term effects by letting test subjects wear the rehabilitation shoe over an extended period of time, where the severity of the after effects is periodically measured. The method by which long t erm effects would be trained is by letting the test subject wear the shoe on the wea k leg and so limiting the forward motion of the weak leg. This motivates the individual to lengthen the forward distance where the initial heel contact point is. The GEMS also pushes the leg backwards simulating a correct gait. It is hypothesized that p rolonged wearing of the GEMS will have positive after effects

PAGE 30

19 helping asymmetric walking patients adapt a more normal walking gait over a longer period of wearing the GEMS. This application of the GEMS also has a potential to increase muscle impedance thr ough the external perturbations [18] resulting in an altered walking gait. Even if no long term permanent aftereffects are found to be present, the GEMS can still be used to correct asymmetric walking patterns by wearing the shoe on the strong leg and let ting it swing past the stance phase, limiting forward progression, and letting individuals toe off with that leg. As a result both legs will push the individual forward by similar distances, evening out the asymmetric gait. This is important considering t hat post stroke patients do not bring one leg back far enough, causing a limp. The result is the shortening of the stance phase, causing an unsuccessful toe off to efficiently propel them forward.

PAGE 31

20 Figure 8 : Rehabilitation method for asymmetric gait ba sed on exaggeration where the GEMS is worn on lagging leg. Lagging leg is pushed backward motivating individual to perform a healthier toe off

PAGE 32

21 Figure 9 : Rehabilitation method for asymmetric gait based on compensation where the GEMS is worn on h ealthy leg. This method even s out the forward progression between both the healthy and lagging leg, generating a symmetric gait 2.3 : GEMS and Context Awareness automatically account for perturbations to the physical body while preparing and adjusting for such disturbances [9] The concept of a broken escalator effect was analyzed and it was proven that the body relies heavily on visual sensory and visual cues for balancing [4 0] me to believe that

PAGE 33

22 gait and in turn the adaptation of symmetric gait in hemiplegic patients. It is this context awareness that is hypothesized to be an integrating factor in the inability to store the previously described feed forward motion learned in split belt gait manipulation res earch [2] While after effects can be achieved [1, 2] as subjects adapt to the asymmetric treadmill speed and are sat out to walk over ground, the learned gait motion disa ppears within seconds. Although the kinematics of walking on a treadmill and th e act of walking over ground seem identical, the visual cues and kinetics are different. The concept of the GEMS eliminates this problem with context awareness during the gait adaptation process. With the GEMS the human gait is altered so that it is in l ine belt treadmill concept, however during conditioning there is no disconnect between the visual cues during the conditioning process and the visual cues after the adap tation process. 2.4 : Previous GEMS M odel An existing GEMS prototype [1] shown in Figure 10 has been developed which successfully generated the desired backward motion simulating a split belt treadmill with a 2:1 track velocity ratio. However, la rge variations from step to step were observed, which is hypothesized to prevent the user from fully adapting to the motion.

PAGE 34

23 Figure 10 : Initial existing GEMS prototype [1] ller, which when a downward force is applied the wheel will cause a backward motion. Figure 11 shows this geared rack in closer detail.

PAGE 35

24 Figure 11 : Rear mechanis [1] The middle roller is coupled with a rail moving forward as the shoe moves backwards, providing a constant two point contact between the shoe and the ground until toe off is initiated. Figur e 12 shows the k inematics of the shoe as the stance phase is initiated until toe off is complete.

PAGE 36

25 Figure 12 : GEMS motion throughout the stance phase [1] The front surface of the existing shoe consists of a free roller and a rubber surface to increas e friction during toe off. The rubber surface is just like any other shoe surface and is relatively flexible allowing the wearer to bend this surface when pushing oneself forward. ward by a total of 10 in (25 cm) on most steps. The GEMS initial prototype yielded similar results as the 2:1 ratio split belt treadmill, while wearing the shoe. However, only minor aftereffects were observed. Only very short term aftereffects of two step s were noted.

PAGE 37

26 This unreliability is thought to be caused by the large variation in dynamics from step to step using the GEMS, where the user had to consciously go through the motions of walking compared to the reliable and even dynamics of a split belt tr eadmill generating a constant velocity profile. The second GEMS version eliminates this unpredictability by having the ability to regulate backward horizontal motion. performed the motion in a jerky, uncontrolled, and unnatural manner. Also instead of slippery surface. It is assumed that this uncontrolled motion activates the bodies balancin g and recovery reflexes, thus hindering a positive adaptation of an altered backward motion of 10 in (25cm), the walking speed was decreased. A huge limitation of the previ ous GEMS model was that it had little adjustability in backward motion velocity, travel distance, or travel direction. There was also a variation of backward stepping distance in each step observed to be caused in the variation of applied user force and w alking speed. The next section will discuss how these limitations will be overcome in the controlled version of the second version of the GEMS. It proceeds to outline each design aspect of the new GEMS.

PAGE 38

27 Chapter 3 : Design and Assembly While the b road concept of this version of the GEMS stems from the previous version, the new version is a complete redesign utilizing different mechanical concepts. The improved GEMS design aims to smooth out the transitions between phases in a human gait by regulat ing the horizontal backward motion of the foot. This controlled motion makes the redesigned GEMS similar to the foot motion experienced when walking on a split belt treadmill The new redesign still acts in a passive manner in that it This new version of the GEMS utilizes an Archimedean spiral shaped wheel to nergy input. Without restriction this backward motion is sudden and uncontrolled, hence the torque created by the GEMS wheels is reduced through a gear train so that a small magnetic particle brake can determine the magnitude of resistance to backward mot ion of the shoe. The magnetic particle brake is controlled by a microcontroller, which determines the magnitude and timing of resistance through a potentiometer coupled to the gear train and an accelerometer attached to the GEMS frame. While one of the u strapped to the GEMS, the opposite foot is strapped to a raised support platform which has the same dimensional and weight properties as the GEMS.

PAGE 39

28 3.1 : Wheel Shape A vital part of the GEMS design is its wheel shape. The whee ls are designed based on the Archimedean spiral shape shown in Figure 13 Figure 13 : The GEMS wheel shape is designed so that when normal vertical force is applied it behaves as in rolling down a slope The radius changes throughout the rotation angl e, which is essentially akin to rolling down a hill with a uniform wheel, but in this case the slope is attached to the foot and is not part of the ground. The radius, R over a whole rotation of an Archimedean spiral shape is obtained by Equation 1, (1) where the constants b and n constitute the size and shape of the Archimedean spiral. Because of the passive nature of the GEMS, in that it rolls on its own weight due to an asymmetric nature of the wheel, the Arc himedean spiral is utilized in creating a

PAGE 40

29 motion. In this shape the wearer applies a vertical force during the stance phase that can be directly related to the instantane ous horizontal backward reaction force through Equation 2. (2) Where L is the instantaneous perpendicular distance between the wheel center and the ground contact point and R is the distance between the ground and wheel center or axle attachment.

PAGE 41

30 Figure 14 : The GEMS wheel shape redirect s vertical user weight force into a horizontal ground reaction applying a torque at the wheel axle creating forward progression Figure 1 5 : GEMS Wheels shape length, L, and radius, R, parameters

PAGE 42

31 Using the Archimedean spiral wheel shape allows the manip ulation of several customized variables such as the horizontal travel distance (270 wheel sector perimeter), horizontal maximum velocity, overall shoe height, and of course horizontal backward all wheel shape variables and by using a custom wheel shape optimization tool (Appendix C ), a wheel shape of appropriate dimensions was selected. The resulting wheel parameters are shown in Table 1. Table 1: Wheel shape parameters chosen for the new GE MS The average horizontal force exerted by the wheel was found using Equation 3, (3) This design of the shoe also allows the direction of the wheels to be changed so the shoe can similarly provide a forward motion. This allows me to put one shoe on each foot where each f oot generates an opposite horizontal motion. Using two shoes would provide the rehabilitative split motion effects in an environment that most closely resembles walking over ground and would allow the greatest ground motion differential

PAGE 43

32 while providing the same forward progression as walking without the shoes does. However, this thesis work only considers the design, construction, and testing of one As shown in Figure 5 in section 2.1, a maximum 170N (38lb) horizontal force is applied slightly after the person makes heel contact in the backward direction and slightly before a person initiates toe off in the forward direction (assisting). The average backward force of 270N (36lb) exerte d by the wheel shape in the horizontal direction slip ping forward after heel contact. Also the design goal of 6.5in (16.5cm) wheel travel distance is accomplished by the selected wheel shape which has a perimeter of 6.7in (17in). This slight overcompensation is appropriate considering that the wheel does not always ideally touch down at the perimeter beginning. The assembly method to mate the wheel to a 0.25in (0.635cm) diameter steel axle is simply the axle pulled through the wheel center with two set screws 60 apart with one set screw laying on a flat that is machined onto the end of the axle. The side view of this assembly method is shown in Figure 16

PAGE 44

33 Figure 16 : Side view of actual GEMS wheel mated to axle Considering the selected wheel shape, as an 800N (180lb) individual steps on all four wheel in the middle of the stance phase, an average torque of 80.1lb in (9.0N m) is exerted on all four wheels, equating an average of 20.0lb in (2.3N m) at each individual wheel. Bearing that and my goal to reduce shoe weight, the wheels were made out of aluminum. Also, to further reduce wheel weight holes were drilled through the wheel. The option of selecting aluminum for axle material was taken into account, however, reaction forces exerted from the mit er gear set in the drive train were too large and caused the aluminum axle to bend, hence a steel 0.25in (0.635cm) axle was selected. Dimensions and positioning of all four wheels are shown in appendix E

PAGE 45

34 The front and back axle of the GEMS are coupled with a set of sprockets and a chain between the two steel sprockets. These sprockets are pinned to each steel axle. The specifications and dimensioning of these individual components can be shown in appendix D while the global setup of the chain and sprocket axle coupling can be viewed in appendix E The coupling of the front and back axle through a rubber timing belt with an intermediate custom made belt failed because the torque on the wheel was too large, causing the belt to slip. 3.2 : Ge ar Train and Magnetic Particle Brake My new design alleviates the largest deficiency of the previous shoe: the large motion variability generated and the jerkiness during each step. To overcome this limitation, the angular velocity of the wheels and in tur n the movement of the GEMS in the new design can be controlled by a braking system. This braking system consists of a gear train and a 0 24V small magnetic particle brake, which when combined is strong enough to resist any shoe motion as a person is steppi ng on the shoe. A magnetic particle brake is essentially a voltage actuated clutch; the more voltage is applied, the harder two internal plates push together impeding shaft rotation. An electric motor capable of generating the same necessary movements wo uld require too much power and would not be available within the weight restriction. The front axle is mated to the rear axle with a chain and sprockets. As the wearer applies a vertical downward force onto any or all of the four wheels, the combined torqu e distributed from the rear axle through the gear train and to the magnetic particle brake for resistance.

PAGE 46

35 The maximum torque exerted by the wheels can be estimated using Equation 1 combined with Equation 4, (4) Two sets of Miter gears were used to redirect the gear train for a space efficient fit. Three spur gears with three 4:1 (60 T 15 T) gear reductions each were used to reduce the torque applied by the GEMS wheels by a factor of 64. This gear train design reduces the input torque at a wheel axle from 40lb in (4.5 N m) to 0.625lb in (0.071 N m). A 1 lb in magnetic particle brake was selected to apply resistance to the reduced torque of to 0.625lb in (0.071 N m). The gear train and magnetic particle brake setup is schematically shown in Figure 17 Friction forces and forces from the spring actuated wheel reset mechanism were also taken into account during magnetic particle brake selection. Figure 17 : Side v iew of the g ear t rain of the GEMS including magnetic particle brake Note that I chose to use a brake to resist the natural shoe movement generated by the users weight applied to the wheels of the GEMS, a spring then generates the return force to the whole mechanism.

PAGE 47

36 3.2.1 Magnetic Particle Brake The GEM movement in the horizontal direction. This redirected force yields horizontal GEMS movement by wheel rotation, this rotation is controlled through a gear train coupled to a magnetic particle brake. Figure 18 shows the magnetic particle brake chosen to resist and stop the GEMS horizontal movement. Figure 18 : 1lb in magnetic particle brake utilized in the GEMS design The specification and option catalog page for the magnetic part icle brake is shown in appendix D A magnetic particle brake is essentially a miniature clutch consisting of two parallel plates pressed against each other creating desired friction, which in turn impedes shaft rotation. Considering the options outl brake needs to stop 0.625lb in (0.071N m) of torque coming from the wheel axle through the gear train, the 1lb in magnetic particle brake was selected. Although the later discussed reset mechanism and fri ctional forces impede wheel rotation, a dynamic force of a stepping motion is considered, yielding a higher torque needing to be stopped. Also

PAGE 48

37 the selected brake weighs 1lb (0.45kg), which is a reasonable weight considering the weight criteria for the GEM S is 2.2lb (1 kg). An electric motor was considered to handle the same type of GEMS movement but no electric motor was found small and light enough to control the torque exerted by the GEMS wheels, hence a passive magnetic particle brake was used The magnetic particle brake is powered by a non inverting op amp circuit shown in Figure 19 Figure 19 : Non Inverting op amp circuit used to output 0 24 to the magnetic particle brake Shown in the above Figure is the circuit that amplifies the a 0 5V pulse width modulation (PWM) signal from the microcontroller and outputs a proportional 0 24V signal to the magnetic particle brake, where 0V is no brake resistance and 24V is the full 1lb in (0.11N m) of resistance. The whole electrical circui t including the circuit outlined above can be viewed in A ppendix A

PAGE 49

38 3.2.2 Gear Train The gear train is used to reduce the torque from the wheels to a torque that can be handled by a small magnetic particle brake. The magnetic particle brake selected for implementing impedance on the movements of the GEMS can exert a maximum torque of 1lb in. The gear train reduces the wheel torque from 40lb in (4.5N m) to a torque that can be stopped by the magnetic particle brake. The gear train is made up of three 4:1 gear reductions for a final 64:1 total reduction in torque. Given the 40lb in (4.5N m) input torque on heel contact, the magnetic particle brake receives 0.625lb in (0.071N m). The slight over compensation of gear reduction stems solely from the larg er dynamic gait forces and magnetic particle brake availability options. While irrelevant to the brake selection, it is worth mentioning that at a maximum wheel angular velocity of 10rpm, the angular velocity at the magnetic particle brake is 640rpm, whi The gear train includes two ninety degree Miter gear sets that allow the 60 tooth and 15 tooth spur gears to be rotated in such a way that their flat parts are parallel to the ground. This setup of spur gears gives the most efficient fit inside the GEMS giving the GEMS an optimal height for the components utilized. These three reductions are done with three sets of 60T 15T reductions rotated by two sets of Miter gears. This setup is schematically depicted in Fi gure 17 and the dimensional details are shown in appendix E Actual images of the gear train setup is shown in Figure 20 Considering the large torque that is exerted by the wheel axle and transferred through the gear train, the Miter gear sets and the s pur gears are chosen to be plain carbon steel. During the initial material selection process, plastic Miter gears were

PAGE 50

39 used which failed due to large reaction forces. Although much lighter, another problem with plastic gear sets is that they ar e very hard to mate to metal rotation shafts and wheel axles due to the low yield strength of the plastic. Figure 20 : GEMS gear train Since the GEMS wheels create a large torque, the first Miter gear set is exposed to large reaction forces pushing the Miter gear set apart. These reaction forces push the top Miter gear into the top frame cover and the bottom Miter gear to the side. These large reaction forces require a stiff bracket in order to keep the Miter gear set meshed during the GEMS operation. This bracket was custom made with aluminum and bolted to the bottom of the shoe frame (Figure 21 ). Also, to keep the Miter gear meshed an aluminum plate was mounted on top of the GEMS frame cover so that the Miter gear reaction forces do not push the ro tation shaft upwards, this mounted aluminum plate is shown in Figure 22 Specifications and dimensions of the Miter gear set are shown in appendix D

PAGE 51

40 Figure 21 : Front m iter gear aluminum bracket Figure 22 : Aluminum plate on top of the GEMS frame cover preventing unmashing of m iter gear set The vertical rotation shafts of the spur gears are also made of steel and are held into place by appropriate roller bearings. The gears are pinned with a 0.125in (0.318cm) steel pins to the 0.25in (0.635cm) wheel axles and rotation shafts.

PAGE 52

41 3.3 : Wheel Reset Mechanism In order for the shoe dynamics to be identical every step, the GEMS resets the wheel position using the spring mechanism schematically shown in Figure 23 This reset mechanism consists of two extension springs, nylon strings, a small redirect pulley, and a pulley which is attached to the first gear axle. Figure 23 : Top view of the GEMS showing the wheel reset mechanism. As the first rotation shaft rotates clockwise the extension spri ng set extends As the wearer applies a vertical downward force on the wheels during the stance phase, the rotation of the wheels also causes the pulley to rotate, pulling the two extension springs apart. This potential energy stored in the extension sp rings is released during the swing phase so the wheels are rotated back to their initial position and ready

PAGE 53

42 for a subsequent step. The springs were selected so that their force can overcome all internal friction of the system while having a sufficient str etch length and stiffness. A pulley is mated to the first rotation shaft of the gear train, on top of the first 60T gear as shown in Figure 23 A doubled nylon string is attached to this mated pulley, pulled across a small redirect pulley, and attac hed to a set of extension springs, which in turn are attached to the GEMS frame by a custom aluminum spring bracket. As the GEMS wheels rotates, so does the Miter gear attached to the axle and in turn the first vertical rotation shaft of the gear train wi th the mated metal pulley. The nylon string that is attached to this pulley is directed across a smaller redirect pulley which then pulls a set of extension springs apart. Once the GEMS is lifted off the ground during the swing phase in the gait cycle and the magnetic particle brake is released, the extension springs and so the GEMS wheels quickly return to its initial position. Selecting the extension spring which can overcome this torque is an iterative process that accounts for several factors: Free len gth, maximum extension spring selection, stiffness, force at extension lengths, availability, and the consideration of combinations of springs in parallel and in series. Combined internal static and dynamic frictional forces are hard to predict and so t o correctly account for all the spring force required to rotate the whole mechanism back to its initial position against all frictional and damping forces, a simple setup was used as depicted in Figure 24

PAGE 54

43 Figure 24 : To estimate the spring forc e need to reset the GEMS, a simple setup was used. A string was tied around the circumference of a wheel of the GEMS and weights were added to the string end until the whole mechanism started rotating. Knowing the wheel radius and the weight, a torque of 10.5lb in (1.2N m) on the wheel axle was calculated. This torque is required to overcome all internal static friction and damping of the GEMS and rotate the wheel to its initial position. The positioning and stiffness of the spring is also a function of where the reset pulley is placed in the gear train and what the radius of the reset pulley is. The further down the pulley is placed in the gear train, the less torque is needed to reset the shoe. Placing a reset pulley on the first rotational shaft of the gear train would require 10.5lb in (1.2 N/m), while the torque required to reset the shoe by placing the reset pulley on the second rotation shaft is 2.625lb in (0.30 N/m). Also of course, reducing the radius of the reset pulley yields a larger force required to overcome the torque needed to reset the whole shoe to its initial position. So the position and size of the reset pulley dictates the extension length of the spring, in that as the reset pulley moves further down the gear

PAGE 55

44 train, more rotations are required to reset the wheels and the greater the circumference the reset pulley is, the longer distance the spring is eventually extended. Given these constraints and spring availability, extension springs were placed in parallel with the reset pulley placed on the first rotation shaft of the gear train, again as shown in the schematic above in Figure 23 In order to keep a pretension on the extension spring set before they are pulled apart, the extension springs are held slightly extended with an other nylon spring attached to the frame of the GEMS as shown in Figure 25 Figure 25 : The extension springs are pre tensioned by a nylon string attached to the frame All the components of the wheel reset mechanism play off each other in that as one component changes, it affects the other components. The two parallel springs were selected with a sufficient stretch, a pre tension, and a maximum force and max stretch length so that when incorporated with a sufficiently large reset pulley would adequ ately cover the needed 10.5lb in (1.2N m) torque to completely reset the shoe. Table 2 shows the spring properties of the chosen extension spring.

PAGE 56

45 Table 2 : Extension spring properties To accommodate these springs, a custom made aluminum pulley of 1. 0in (2.54cm) radius was selected and placed on top of the first 60T gear pinned to the first rotational shaft. This aluminum pulley was held into place by two set screws 180 degrees apart, screwed down onto the top of the 60T gear. On one side a second set screw was used as an attachment point for the doubled nylon string ( Figure 26 ). For detailed reset pulley dimension refer to appendix E Figure 26 : Aluminum reset pulley, redirect pulley, and doubled nylon string

PAGE 57

46 A little redirect pulley o f 0.225in (.573cm) radius was designed and custom machined out of aluminum with a steel pin holding this pulley, this pulley was placed in the back upper corner of the GEMS frame. A nylon string was used to pull the extension springs apart. While wires and ropes were either inflexible or too thick, the nylon string was very low friction, strong, and very flexible, however by itself was too weak to withstand the extension spring set force when extended to the maximum, hence, the nylon string was doubled, cutting the tension in each chord by half. The nylon string was then in turn attached to the extension springs using a specially made aluminum bracket machined in a way to where the both the extension springs neatly hooked on it and the nylon string ti ed around it. The bracket that connects the GEMS frame to the extension spring is again machined out of a small block of aluminum and screwed into the back and the side of the GEMS. Figure 27 : Spring to nylon string bracket (left) and spring to frame bracket (right)

PAGE 58

47 A constant force spring was considered early in the GEMS design because of its attractive ability to exert constant force onto the system, this idea however was proven complicated to implement, and reduced the reliability of the GEMS sho e. The constant force spring needed a delicate guide and attachments in order for the spring not to coil up violently. All dimensions and drawing concerning the wheel reset mechanism can be viewed in appendix E 3.4 : Electronics The GEMS varies t he motion resistance through the magnetic particle brake using an op amp circuit in conjunction with a BS2p24 microprocessor (Figure 28 ). Depending on what point in the gait cycle the wearer is, variable resistance is applied to the GEMS. Instances in th e gait cycle are identified by using a rotational potentiometer and an accelerometer. While the potentiometer recognizes the wheel rotation, the accelerometer recognizes when heel contact and toe off occur. Heel contact is measured by the jerk applied to the wheel and frame of the GEMS, and toe off is determined when the shoe has tipped forward by 30 degrees. In order to easily reprogram the on board microprocessor, an external RS232 connection was mounted outside the side of the GEMS. All electronics ar e powered by a small battery pack, which the user wears on their hip. The electrical diagram for the combined GEMS electronics is shown in appendix A

PAGE 59

48 3.4.1 Microcontroller A Parallax BS2p24 microcontroller was used to collect sensory information fro m a potentiometer and an accelerometer, and control the timing and resistance of a magnetic particle brake and so in turn the GEMS movement. The specification and electrical diagram for the selected microcontroller are found in appendix D Figure 2 8: Parallax BS2p24 microcontroller integrated into the GEMS This chosen microcontroller speed is 20 MHz which guarantees that acceleration instances are well captured while it while analyzing the potentiometer input and outputting appropriate resistive torque through the magnetic particle brake. Programming the microcontroller was achieved by a RS232 cable that connected a desktop computer to the microcontroller. To minimize hassle the female connection of the RS232 connection coming from the microc ontroller was pulled through the side of the GEMS (Figure 28 ).

PAGE 60

49 The programming of the microcontroller for the GEMS was very straight forward and needed to be simple to enhance runtime speed. It included the following: Scanning of the potentiometer for wh eel rotational position Scanning of accelerometer for shoe angle and sudden acceleration spike Determining where in the gait cycle the GEMS is located Dictating magnitude and timing of the resistance exerted by the magnetic particle brake Outputting of vo ltage dictating how much resistance the magnetic particle brake exerts Initially proportional and derivative controls were to be programmed into the microcontroller but the result was unreliable and further investigation in code depended controls is needed The final code assigned a set resistance to the magnetic particle brake at key instances of the gait. This code can be viewed in appendix F The BS2p24 was integrated into a distinctive circuit board drawn up in PCB Artist and printed by 4PCB.com comp any. This small circuit board schematic as it was printed is shown in Figure 29

PAGE 61

50 Figure 29 : Circuit board for GEMS The circuit board is mounted on a custom acrylic bracket which is screwed to the side of the GEMS right above the chain which co nnects the two axles. This acrylic bracket was produced by a VLS4.60 Universal Laser Systems laser cutter using 0.125in (0.32cm) clear acrylic. 3.4.2 Accelerometer During the gait cycle there are some distinct instances that are important to the GEMS: Heel contact and Toe off. Both of these instances are borders for the stance phase and the swing phase and can be identified using a accelerometer. Heel contact and so the initiation of the stance phase is recognized both, by a sudden acceleration or jerk when the GEMS wheel hits the ground, and a slight tilt in the GEMS shoe as the foot is approaching heel contact. Toe off and so the initiation of the swing phase was

PAGE 62

51 recognized by a slight forward tilt of the foot. In these instances the acceleromet er readings are recognized by the microcontroller and appropriate resistances are applied by the magnetic particle brake. The Parallax Memsic 2125 Dual Axis accelerometer (appendix D ) was chosen to accomplish this task. In convenience this acceleromet er is very easily complimented to the BS2p24 microcontroller due to the fact that it was produced for this type of microcontroller which reads a pulse width modulation (PWM) 0.5V input. Its power requirement is 3.3VDC to 5.0VDC, and is obtained by tapping into the BS2p24, which 12VDC down to 5VDC. Due to tight space inside the GEMS the accelerometer was positioned in the top allows ob servation of the acceleration change during heel contact. The positioning of the accelerometer is shown in Figure 30 Figure 30 : Position of accelerometer inside the GEMS

PAGE 63

52 Also the schematic for the accelerometer as it relates to the microcontroller is shown in appendix A The circuit board used on which the accelerometer lies was printed by using PCB Artist provided by 4PCB.com and is depicted in Figure 31 Figure 31 : Circuit board for the accelerometer 3.4.3 Potentiometer The rotational posit ion of the GEMS wheels are also important to differentiate between the swing phase, when the magnetic particle brake applies no resistance, and the stance phase, when the magnetic particle brake applies a constant resistance impeding sudden GEMS motion. T he potentiometer is attached to an aluminum bracket on the side of the GEMS next to the second rotation shaft of the gear train and right underneath the second 60T spur gear. The setup of the potentiometer is shown in Figure 32

PAGE 64

53 Figure 32 : Poten tiometer attached to the gear train shaft A small aluminum timing belt pulley is attached to the potentiometer and an identical timing belt pulley is attached to the second rotation shaft of the gear train. In order to eliminate slippage and obtain an ac curate reading, a small timing belt connects the two aluminum timing belt pulleys. Due to the lack of space within the GEMS shoe the potentiometer had to be positioned away from moving parts of the shoe such as the extension springs used for the wheel r eset mechanism (Section 3.3, Section 4.5). A custom machined bracket holds the potentiometer away from the GEMS wall so that when the extension spring is extended it does not interfere with the potentiometer. This aluminum bracket is screwed together by a set of screws to the GEMS wall. Specification for the potentiometer component are found in appendix A electrical diagram relating the potentiometer to the microcontroller is found in appendix A and a GEMS layout which includes the potentiometer is found in appendix E

PAGE 65

54 3.4.4 Battery Pack Electrical components in the GEMS require different voltages. The magnetic particle brake operates at 0 24VDC and the microcontroller operates at 5 12VDC (The accelerometer can tap into a regulated 5VDC pin on the microcontroller), and the op amp also requires a 3VDC supply. The electrical requirements for the magnetic particle brake, microcontroller, and accelerometer are shown in Table 3 Table 3 : Power requirement for GEMS components Considering the pr eviously listed GEMS criteria of an uninterrupted operation criteria of 1.5 hours and the given elec trical requirements outlined in Table 3 a battery pack battery pack consisting of three separate power supplies was custom made and is shown in Figu re 33

PAGE 66

55 Figure 33 : GEMS battery pack A 7.5VDC power supply was simply made from five separate 1.5VDC AA batteries connected by AA battery holder. The same AA battery holder was tweaked to hold a separate two 1.5VDC AA batteries to create a 3VDC vol tage. While the 24VDC supply running to the op amp was made with two 12VDC 500mAH rechargeable Ni Cd batteries connected together in parallel to yield 24VDC. Knowing the power requirements for each component, the 24VDC battery can be continuously be ope rational for 5.5 hours, 7.5VDC for 46 hours, and 3VDC for 50 hour, hence the time the GEMS is fully operational is 5.5 hours, which is well over the set criteria.

PAGE 67

56 3.5 : GEMS Frame The symmetric frame for the GEMS was chosen in a way so it could be used in both directions as the wheels can be repositioned. While the GEMS frame itself is important to the general operation of the GEMS, it was designed around other design requirements, which included the size of the gear train, the size and shape of th e wheels, and the size of the magnetic particle brake. The simplest shape was chosen for the frame design, a rectangular box with rounded rubber pieces at the lower front and back corners. These rubber pieces support a stable heel contact and toe off dur ing a gait cycle. It is made out of light and strong 0.1875in (0.5cm) fiberglass and held together with various aluminum brackets. Two rubber pieces were added to the lower front and back corner of the shoe so the wearer could effectively create a solid he el contact and toe off. Fiberglass with 0.1875in (0.5cm) thick ness was selected for the frame material. A Pugh analysis for the selection of this material can be reviewed in appendix B This machinabilty. The tensile strength for the fiberglass used for the GEMS frame is 30ksi (206MPa) lengthwise and 7ksi (48MPa) crosswise It has an impact strength of 25 ft lbs/in lengthwise and 4 ft lbs/in crosswise. The thick rubber pieces that replaced lower front and back corners were also chosen to be of 0.1875in (0.4762cm) thickness. The whole frame is held together with 0.0625in (0.1587cm) aluminum L brackets at the corners of the frame. The L brackets are held into place by screwing them into the frame.

PAGE 68

57 Bearings holding wheel axles and rotation shafts for the drive axles were press fit into the frame at appropriate positions. These bearings accommodate a 0.25in (0.635cm) shaft for the wheel axle and for rotat ion shafts in the gear train. All dimensions and specifications of the frame and its components can be reviewed in appendix E Figure 34 : GEMS Frame and bottom of frame Cover 3.6 : Shoe Straps the top of the GEMS. Identical to the first GEMS design, these straps were designed after a traditional sandal design (Figure 35) rigidly su pporting the whole foot with minimal straps, this ensured minimal movement of the foot relative to the GEMS. Velc ro straps were utilized for

PAGE 69

58 Figure 35 : The GEMS straps are designed after traditional sandals 3.7 : Opposite Leg Support Platform porting platform of equal height and weight for the opposite foot was constructed. The support platform was designed to be the exact same dimensions, weight, and fastening style to eliminate any unnecessary asymmetries. This platform was made with a thic k rubber sole to maximize friction and stepping smoothness. To match the weight difference, lead weights were glued to a stand in the middle of the support platform. The dimensional specification of the opposite leg support platform can be viewed in appe ndix E

PAGE 70

59 Figure 36 : Platform used to compensate for the height of the GEMS 3.8 : CAD Model Verification A SolidWorks 3D CAD model was created and altered as the GEMS design progressed. This CAD model was used for spacial and dimensional predic tive purposes as well as the creation of technical drawings conveniently used for custom manufacture of shoe components. This model made it very simple to create any changes, troubleshoot any dimensional issues, and predict issues for any proposed componen ts or changes. Figure 37 shows snapshots of the final SolidWorks 3D CAD model.

PAGE 71

60 Figure 37 : GEMS SolidWorks 3D CAD model

PAGE 72

61 Chapter 4 : Results The assembled and functioning GEMS is shown in Figure 38 and in Figure 39. Although not a complete motion analysis was performed on the GEMS, the completed GEMS was evaluated by wearing it for walking on it three to four steps at a time. By doing this an effective horizontal push length was measured and the GEMS movements were observed. An effective way to describe the resulting GEMS is to compare it to the initial criteria outlined in chapter 1. Figure 38

PAGE 73

62 Figure 39 : Interior of complete GEMS 1. Total weight : The final GEMS has a total weight of 4.5lb (2 kg). Majority of this weight is contributed by the magnetic particle brake, spur gears and miter gears in the gear train, and the fiberglass frame. Although this new GEMS design alleviates the major problem of the last GEMS in that it moves smooth and controllable, it is too heavy and requires optimization in that sense. For reference, an average everyday shoe weighs in the range of two to three pounds 2. Strength : Although the new GEMS becomes unreliable when a person of 115 kg proceeds to walk on it, i t succeeds in withholding all static and dynamic forces exerted by a 90 kg person during walking. 3. Generated motion: The finished GEMS generates a backwards motion of 15.2 cm each step. This backward distance of 15.2 cm successfully mimics half of the full backward motion generated by split belt gait rehabilitation research. Placing a similar GEMS design with wheels generating a forward progression

PAGE 74

63 on the opposite foot instead of the support platform would successfully generate a relative 30.4 cm. This mot ion is smooth and controlled and is depicted in Figure 40. Figure 40 : GEMS stepping motion. Not e that each floor tile is 12in (30.5cm) 4. Consistent motion : While the first GEMS prototype worked well in that it generated a backward motion, it had gr eat variability in each step. Not only could each step be shorter or longer, the backward velocity at which the foot was pushed back could be different at each step and with each person. This ion reflexes and seems unnatural. This problem was solved with the new GEMS design in that reduced shoe movement variability by consistently using the same resistance on each step with a controlled magnetic particle brake. While initial tests looked prom ising in the sense of reducing GEMS variability during each step, further

PAGE 75

64 gait analysis with the GEMS is required to draw a stable conclusion about the long term performance of the shoe. 5. Portability : The new GEMS, as the old GEMS, is completely portable a nd can be worn on any hard surface such as carpet, concrete, or floor tiles. It is also 6. Time to recharge : The new GEMS prototype utilizes a battery that exceeds prior design criteria of lasting 1.5 h ours, it is estimated to last 4.5 hours. However, the new GEMS was not completely tested over that time span and further investigation is necessary. 7. Size (height) : The height of the new GEMS does not meet the proposed height criteria of 2.5in and has a fin al height of 3.8in. The major components that cause this height are the gear train and the wheels. Placing the gear train on the inside of the frame naturally yields a frame height equal the gear train height while the initial height of the wheels accoun t for the rest of the height. 8. Size (width) : Since none of the components significantly influence the shoe width, the GEMS has a 4.375in frame width and a 5.375in width including wheels. This GEMS prototype meets the proposed width criteria.

PAGE 76

65 9. Size (length) : The length criteria is met by the new GEMS design with a length of 10.375in. While it satisfies the criteria of an average sized tennis shoe, the length is the result of the gear train and the magnetic particle brake imbedded inside the GE MS frame. 10. Cost effectiveness : The finished new GEMS and the previous GEMS do not currently meet this criteria in that both are custom made designs made from various custom components and include extensive manual labor. Neither shoe was optimized for manuf acture. 11. Shoe Progression: The GEMS successfully utilizes four wheels in the shape of redirects it to a backward motion. This wheel shape should allow a total push distance of the shoe of at least 6.5in (15cm). 12. Straps: The same shoe strapping technique that was used in the previous GEMS was utilized in the new GEMS design allowing minimal movement between the 13. Opposite Foot Support Platform: Because of the height difference between he GEMS and the opposite foot which sits on the ground, a foot platform of the same height and weight was constructed. It successfully mimics the same height, width, length, and weight of the second GEMS prototype.

PAGE 77

66 Table 4 shows the s ummery of the above criteria and how the first and second GEMS prototypes have or have not met them. Table 4: New and o ld GEMS compared by criteria

PAGE 78

67 Chapter 5 : Future Work Even though many drawbacks of the previous design were corrected there ar e numerous improvement opportunities in the design. Such improvements and optimization include in the following areas: Material selection This is a big category for improvement from the new GEMS design. While many current material design decisions are reasonable, more investigation in how different materials can benefit with different components is necessary. Wheel transition Adding a middle point of contact between the front and back axle is needed to alleviate an abrupt transition between the two po ints of contact with the ground Balancing of forces Balancing the forces between the gear train, magnetic particle brake, wheel shape, frictional forces, and reset mechanism is also a major category of improvement and optimization. These four GEMS compo nents very much affect each other and determine the properties of each other. For instance, the shape of the wheel determines how much torque at what instance of the gait cycle is generated while the magnetic particle brake and the gear train is to be des igned in such a way to where they can resist this generated torque. Wheel Shape The Archimedean spiral wheel shape alone is an interesting and agile aspect of the GEMS and is open to a detailed analysis, possibly resulting in shaping the wheel in such a way where desired horizontal forces are produced at

PAGE 79

68 specific instances of rotation. Further investigation on how to utilize a more useful wheel shape exerting desired forces at specific instances is needed. Frame design The current GEMS used a standard rectangular boxed frame design made out of fiberglass. The frame can easily be optimized not only in shape but also in weight by carefully designing a skeletal type shape frame with various material components. Gear train layout The gear train is a vita l part of the current GEMS design and needs further attention in how it is laid out. Different types of gear train layouts can be considered, including one where the gear train sits partially outside the shoe shape. Furthermore, the GEMS as developed is s ufficient enough for test trials revealing if in fact this design can affect more positive gait altering effects than the previous version and comparable to previous split belt research.

PAGE 80

69 Chapter 6 : Conclusions I successfully designed and construc ted a functioning gait enhancing mobile shoe (GEMS). Although the previous version of the GEMS effectively showed some after s foot back in a sudden motion analogous to slipping on ice. This type of sudden motion triggers instincts, thus producing an unnatural feel. This unnatural motion was greatly reduced in this version of the motion controlled GEMS model. My improved model is easily adjustable to different horizontal push length, force, speed and direction by simply adjusting the wheel size, wheel shape, and magnetic particle brake resistance. This adjustability in behavior of the G EMS makes testing for various situations possible. While this new design of the GEMS is promising and is a step forward from the previous GEMS design, room for optimization are plentiful. These optimizations include, but are not limited to, material selec tion, wheel shape design, control shoe resistance design, or reset mechanism design. Furthermore, during the design and assembly process of this new GEMS many practical and technical missteps were taken from which valuable GEMS design skills were acquire d for proceeding versions of the GEMS. These missteps range anywhere from machining practices to design approach strategies.

PAGE 81

70 All in all the new GEMS design is successful in satisfying most of the initially proposed design criteria and giving more insight in its design process.

PAGE 82

71 References [1] A. Groot, R. Decker, K. Reed. Gait Enhancing Mobile Shoe (GEMS) for Rehabilitation. Third Erophaptics Conference and Symposium on Haptic Interfaces for Virtual Enviroment and Teleoperator Systems. March 18 20, 200 9. [2] D. Reisman, R. Wityk, and A. Bastian. Split belt treadmill walking adaptation in post stroke hemiparesis. J. Neurologic Physical Therapy, 29:196, 2005. [3] T. Lam, M. Anderschitz, and V. Dietz. Contribution of feedback and feedforward strategie s to locomotor adaptations. J. Neurophysiol, 95:766 773, 2006. [4] S. Morton and A. Bastian. Cerebellar Contributions to Locomotor Adaptations during Splitbelt Treadmill Walking. J. Neurosci., 26(36):9107 9116, 2006. [5] Lam T, Wolstenholme C, and Yang JF. How do infants adapt to loading of the limb during the swing phase of stepping? J Neurophysiol 89: 1920 1928, 2003. [6] R. Grasso, L. Bianchi, and F. Lacquaniti. Motor Patterns for Human Gait: Backward Versus Forward Locomotion. J. Neurophysiol, 80 (4):1868 1885, 1998 [7] J. Perry. 1992. Gait Analysis: Normal and Pathological Function. Thorofare, NJ: Slack Inc [8] P. Perkins and M. Wilson. Slip resistance testing of shoes new developments. Ergonomics, 26(1):73 82, 1983. [9] Whittle, M. W.,199 1. Gait Analysis: An introduction. Butterworth Heinemann Ltd, Oxford. [10] R. Keamey. Context dependence of intrinsic and reflex contributions to ankle stiffness. In Proc. of 1st International IEEE EMBS Conf. on Neural Engineering, pages 434 437, March 2 003. [11] A. Gordon, G. Westling, K. Cole, and R. Johansson. Memory representations underlying motor commands used during manipulation of common and novel objects. J. Neurophysiol, 69(6):1789 1796, 1993. [12] J. Krakauer, M. Ghilardi, and G. Ghez. Inde pendent learning of internal models for kinematic and dynamic control of reaching. Nature Neuroscience, 2:1026 1031, 1999.

PAGE 83

72 [13] Lou JS and Bloedel JR. A new conditioning paradigm: conditioned limb movements in locomoting decerebrate ferrets. s 84: 185 19 0, 1988. [14] Hodgson JA, Roy RR, de Leon R, Dobkin B, and Edgerton VR. Can the mammalian lumbar spinal cord learn a motor task? Med Sci Sports Exerc 26: 1491 1497, 1994. [15] Timoszyk WK, De Leon RD, London N, Roy RR, Edgerton VR, and Reinkensmeyer DJ The rat lumbosacral spinal cord adapts to robotic loading applied during stance. J Neurophysiol 88: 3108 3117, 2002 [16] Arshavsky YI, Orlovsky GN, Panchin YV, Roberts A, Soffe SR (1993) Neuronal control of swimming locomotion: analysis of the pteropod mollusk Clione and embryos of the amphibian Xenopus Trends Neurosci 16:227 233. [17] Yanagihara D, Kondo I. Nitric oxide plays a key role in adaptive control of locomotion in cat. Proc Natl Acad Sci USA 1996; 93: 13292 7. [18] D. Reinkensmeyer, J. Em ken, and S. Cramer. Robotics, motor learning, and neurologic recovery. Annu. Rev. Biomed. Eng., 6:497 525, 2004. [19] H. Schmidt, S. Hesse, R. Bernhardt, and J. Kr¨uger. HapticWalker a novel haptic foot device. ACM Trans. on Applied Perception, 2(2):1 66 180, 2005. [20] M. Otis, C. du Tremblay, F. De Rainville, M. Mokhtari, D. Laurendeau, and C. Gosselin. Hybrid control with multi contact interactions for 6 DOF haptic foot platform on a cable driven locomotion interface. In Symposium on Haptic Interfa ces for Virtual Environments and Teleoperator Systems, pages 161 168, March 2008. [21] D. Reisman, R. Wityk, K. Silver, and A. Bastian. Locomotor adaptation on a split belt treadmill can improve walking symmetry post stroke. Brain, 130(7):1861 1872, 2007 [22] G. I. Turnbull, J. C. Wall. Long term changes in hemiplegic gait. Gait & Posture. Vol. 3: 258 261, December 1995 [23] Bohannon R W, Andrews A W, Smith M B. Rehabilitation goals of patients with hemiplegia. Inl J Rehabil Res 1988; II:181 183. [24 ] M. H. Woollacott. Posture and Gait from Newborn to Elderly. Posture and Gait: Proceeding of the 9 th Int. Symp on Postural and Gait Research, Marseille, France, 29 May 1 June 1988 [25] M. H. Granat, D.J. Maxwell, C.J. Bosch, A.C.B. Ferguson, K.R. Lee s and J.C. Barbenel. A body worn gait analysis system for evaluating hemiplegic gait. Med. Eng. Phys. Vol. 17, No. 5, pp. 390 394, 1995

PAGE 84

73 [26] Kirtiey, C., Whittle, M.W. and Jefferson, R.J. Influence of walking speed on gait parameters. J Biomed Eng 1985, 7,282 288 [27] Georgia Southern University. Vicon Camera [Online] Available http://chhs.georgiasouthern.edu/hk/images/ath pics/current students/ViconCamera.JPG. March 14, 2011 [28] Lab Safety Supply. Conductive Mat. [Online] Available http://www.labsafety.com/images/xl/WEARWELL Electrically Conductive Mats LSS _i_LBV31604_01.jpg March 14, 2011 [29] Wall, J.C., Dhanendran, M. and Klenerman, L. A method of measuring the temporal/distance factors of gait. Biomed Engng 1976, 11, 409 412 [30] Wall, J.C., Charteris, J. and Hoare, J.W. An automated on line system for measuring the temporal patterns of foot/floor contact J Med Eng Tech 1978, 2, 187 1 90 [31] Wall JC, Turnbull GI. Gait asymmetries in residual hemiplegia. Arch Phys Med Rehabil 67:550 553, 1986 [32] Lin PY, Yang YR, Cheng SJ, Wang RY. The relation between ankle impairments and gait velocity andb symmetry in people with stroke. Arch Ph ys Med Rehabil 87:562 568, 2006 [33] Plotnik M, Giladi N, Balash Y, Peretz C, Hausdorff JM. Is freezing of gait in 663, 2005 [34] Marder E, Calabrese RL. Principles of rhythmic motor pattern generation. Physiol Rev 76:687 717, 1996 [35] Grillner S. Control of locomotion in bipeds, tetrapods and fish. In: Brookhardt J, Mountcastle G (eds) Handbook of physiology. American Physiological Society, Bethesda, pp 1179 1236, 1981 [36] Yang YR, Chen YC, Lee CS, Cheng SJ, Wang RY, Dualtask related gait changes in individuals with stroke. Gait Posture [Epub ahead of print], 2006 [37] Jensen L, Prokop T, Dietz V (1998) Adaptational effects during human split belt walking: influence of af ferent input. Exp Brain Res 118:126 130 [38] M. Gaviria et al. Plantar dynamics of hemiplegic gait: A Methodological Approach. Gait & Posture 297 307, 1996 [39] Monco Medical. Split Belt Treadmill. [Online] Available http://www.alibaba.com/ productfree/262579915/Medical_Research_Treadmills /showimage. html March, 14, 2011.

PAGE 85

74 [40] K. L. Bunday, A. M. Bronstein. Visuo Vestibular Influences involve d in the t aftereffect. Gait & Posture, Volume 21, June 2005, Page S23. [41] Wikipedia. Vicon Camera [Online] Available http://uploa d.wikimedia.org/wikipedia/commons/4/42/Copenhagen_Metro_escalators.jpg March 22 2011

PAGE 86

75 Appendices

PAGE 87

76 Appendix A : GEMS Electrical Diagram Figure A1: GEMS Electrical Diagram

PAGE 88

77 Appendix B : Pugh Analysis: GEMS Fram e Material Figure B1: Pugh Analysis: GEMS f rame m aterial

PAGE 89

78 Appendix C: Archimedean Spiral W heel Shape Selection Tool Figure C1: Archimedean Spiral wheel shape selection tool

PAGE 90

79 Appendix D: Specificatio n Sheets Figure D1: BS2p24 module schematic

PAGE 91

80 Appendix D: (Continued) Table D1: BS2p24 Microcontroller s pecification

PAGE 92

81 Appendix D: (Continued) Figure D 2 : Accelerometer s pecifications

PAGE 93

82 Appendix D: (Continued) Figure D 3 : Magnetic particle brake specifications

PAGE 94

83 Appendix D: (Continued) Figure D 4 : Op Amp specifications

PAGE 95

84 Appendix D: (Continued) Figure D 5 : Sprocket connecting front and back GEMS axle

PAGE 96

85 Appendix D: (Continued) Figure D 6 : Chain connecting front and back GEMS axle

PAGE 97

86 Ap pendix D: (Continued) Figure D 7 : Miter gear in GEMS gear train

PAGE 98

87 Appendix E: Dimensions and Drawings Figure E 1 : GEMS assembly

PAGE 99

88 Appendix E: (Continued) Figure E2: Open GEMS assembly top view

PAGE 100

89 Appendix E: (Continued ) Figure E3: GEMS wheel

PAGE 101

90 Appendix E: (Continued) Figure E4: GEMS frame bottom cover

PAGE 102

91 Appendix E: (Continued) Figure E5: GEMS frame side

PAGE 103

92 Appendix E: (Continued) Figure E6: GEMS frame top cover

PAGE 104

93 Appendix E: (Continued) Figure E7: Sixty tooth gear in gear train

PAGE 105

94 Appendix E: (Continued) Figure E 8 : Miter gear bracket

PAGE 106

95 Appendix E: (Continued) Figure E9: Potentiometer bracket

PAGE 107

96 Appendix E: (Continued) Figure E 10 : Reset mechanism redirect pulley bracket

PAGE 108

97 Appendix E: (Continued) Figure E 11 : Small reset mechanism pulley

PAGE 109

98 Appendix F : Microcontroller Code 'ISMET HANDZIC 'Gait Enhancing Mobile Shoe (GEMS) microcontroller PBASIC program 'University of South Florida '2009 2010 {$STAMP BS2p} {$PBASIC 2.5} CycAdj CON $09E 'cycle adjustment for 1 ms Cycles CON 60 'FIX PWM VALUE --RIGHT NOW: GOES FROM 1.9V to 1.7V with pot co mmands Brake_T VAR Word 'Brake torque (0 255 = 0~4.8V = 0~21V) Brake_T_Byte VAR Byte x VAR Word 'Accelerometer X direction phase VAR Bit 'Differentiates between wheel rolling or wheel reseting x_old VAR Word dX VAR Word dX_max VAR Word Pot_RA VAR Word 'Potentiometer rotational angle value Pot_RA_D VAR Word Pot_RA_Old VAR Word Pot_RA_D_Old VAR Word Vel_A VAR Word Vel_D VAR Word 'P VAR Word 'D VAR Word RC PIN 8 'Assign pin 7 as RC Time pin for Potentiometer Brake_T = 255 'Brake start off torque (max torque) Pot_RA_D = 630 x=6700 x_old = 6700 dX_max = 0 phase = 1 Main: DO 'POTENTIOMETER READING HIGH RC 'charge the cap PAUSE 1 'for 1 ms RC TIME RC, 1, Pot_RA 'measure RC discharge time (Pin, 0 or 1 logic, variable) IF(Pot_RA < 430) THEN phase = 1 'swing phase

PAGE 110

99 Appendix F : (Continued) ENDIF IF(Pot_RA > 580) THEN phase = 0 'stance phase ENDIF IF(phase = 1) THEN Brake_T = 0 ELSE Brake_T = 50 ENDIF Brake: 'BRAKE TORQUE OUTPUT IF(Brake_T > 255) THEN Brake_T_Byte = 255 ELSE Brake_T_Byte = Brake_T ENDIF PWM 15, Brake_T_Byte, (Cycles */ CycAdj) 'PWM Pin, Duty cycle, cyles 'Adjust capasitor charging time (5*R*C) if needed 'Adjust Cycles value if needed LOOP END